Item: March wet avalanche prediction at bridger bowl ski area, montana
Title: March wet avalanche prediction at bridger bowl ski area, montana
Proceedings: Proceedings of the 2004 International Snow Science Workshop, Jackson Hole, Wyoming
Authors: J. M. Romig, S. G. Custer, K. Birkeland, and W. W. Locke, Earth Sciences Department, Montana State University, Bozeman, MT
Abstract: Few avalanche forecast models are tailored specifically for wet avalanche forecasting. Bridger Bowl (intermountain climate) is a good area to develop a wet avalanche probability model. The primary archived data consists of eight variables. The archived data for March from 1968 to 2001 (1996 data unavailable) were used to develop 68 predictor variables related to temperature, snowpack settlement, and precipitation. The original dataset was divided into days with snowfall in the past 48 hours (new snow) and days without (old snow). There were 33 significant old snow variables and 22 significant new snow variables. Six variables are common to both old and new snow. The best predictor variables for old and new snow are different. The variables were analyzed with binomial logistic regression to produce probability models for old snow and for new snow wet avalanche conditions. The old snow model uses the prediction day minimum temperature and the two-day change in total snow depth as predictor variables and has a 89% overall success rate. However, the majority of this success is due to correct prediction of days without wet avalanches (96% of all correct predictions). The new snow model uses the prediction day minimum temperature and three-day cumulative new snow water equivalent as predictor variables, but is less useful. The models are applicable only to Bridger Bowl. The numerical forecast models can be used as one of the tools in the forecasting toolbox but limited data and complexity of process require that the decisions about closure remain in the hands of the ski patrol.
Language of Article:
Keywords: wet snow, avalanche, probability, forecast, model
Digital Abstract Not Available